R Notebook

library(ggplot2)
library(ggpubr)
library(CDM)
library(boot)
library(tidyverse)
library(dummy)
library(stringi)
library(stringr)

rm(list = ls())

Q_from_book <- read_csv("data/final_result_similar.csv")

Parsed with column specification:
cols(
  Question = col_character(),
  Option1 = col_character(),
  Option2 = col_character(),
  Option3 = col_character(),
  Option4 = col_character(),
  Answer = col_character(),
  `Learning Objective` = col_character(),
  Topic = col_character(),
  `Difficulty Level` = col_character(),
  `Skill Level` = col_character(),
  `APA Learning Objective` = col_character()
)

glimpse(Q_from_book)

Observations: 1,006
Variables: 11
$ Question                 <chr> "Which of the following is an example of social influence?", "Which of the following...
$ Option1                  <chr> "a. You feel guilty because you lied to your trusting professor about your assignmen...
$ Option2                  <chr> "b. When you get hungry, you have trouble concentrating.", "b. Ramona works hard in ...
$ Option3                  <chr> "c. You didn\u0092t do well on the test because you stayed up all night cramming.", ...
$ Option4                  <chr> "d. You almost fall asleep at the wheel, so you pull off the road to take a short na...
$ Answer                   <chr> "A", "A", "D", "C", "A", "C", "C", "B", "D", "D", "C", "B", "C", "A", "B", "C", "D",...
$ `Learning Objective`     <chr> "1.1 Define social psychology and distinguish it from other disciplines.", "1.1 Defi...
$ Topic                    <chr> "Defining Social Psychology", "Defining Social Psychology", "Defining Social Psychol...
$ `Difficulty Level`       <chr> "Moderate", "Moderate", "Moderate", "Moderate", "Easy", "Moderate", "Moderate", "Mod...
$ `Skill Level`            <chr> "Understand the Concepts", "Understand the Concepts", "Understand the Concepts", "Un...
$ `APA Learning Objective` <chr> "1.1 Describe key concepts, principles, and overarching themes in psychology.", "1.1...

Q_from_book

NA

learning_obj <- Q_from_book %>% distinct(`Learning Objective`) %>% mutate(lo_id = row_number())
learning_obj

NA

Q_pre <- Q_from_book %>% inner_join(learning_obj) %>% select(Question, `Learning Objective`, lo_id) %>% mutate(temp = str_trim(str_replace_all(Question, "_|\\.", "")))

Joining, by = "Learning Objective"

Q_pre

NA

Q.distinct.id <- read_csv("data/Q_distinct_id.csv") 

Parsed with column specification:
cols(
  value = col_character(),
  Q_UNIQUE_ID = col_double()
)

Q.distinct.id <- Q.distinct.id %>% mutate(temp = str_trim(str_replace_all(value, "_|\\.", "")))
Q.distinct.id

Q_pre %>% inner_join(Q.distinct.id, by = "temp") 

NA

Q <- Q_pre %>% inner_join(Q.distinct.id, by = "temp") %>% distinct(Q_UNIQUE_ID, lo_id) %>% arrange(Q_UNIQUE_ID) %>%
  mutate(present = 1) %>%
  
  spread(key = "lo_id", value = "present")

Q

Q %>% write_csv("data\\Q.csv")

Q_pre %>% anti_join(Q.distinct.id) %>% distinct(Question, temp) %>% select(-temp ) %>% write_csv("missing.csv")

Joining, by = "temp"